Magnetic Barkhausen Noise Sensor: A Comprehensive Review of Recent Advances in Non-Destructive Testing and Material Characterization
Polyxeni Vourna, Pinelopi P. Falara, Aphrodite Ktena, Evangelos Hristoforou, Nikolaos D. Papadopoulos
Abstract
Magnetic Barkhausen noise (MBN) represents a powerful non-destructive testing and material characterization methodology enabling quantitative assessment of microstructural features, mechanical properties, and stress states in ferromagnetic materials. This comprehensive review synthesizes recent advances spanning theoretical foundations, sensor design, signal processing methodologies, and industrial applications. The physical basis rooted in domain wall dynamics and statistical mechanics provides rigorous frameworks for interpreting MBN signals in terms of grain structure, dislocation density, phase composition, and residual stress. Contemporary instrumentation innovations including miniaturized sensors, multi-parameter systems, and high-entropy alloy cores enable measurements in challenging environments. Advanced signal processing techniques-encompassing time-domain analysis, frequency-domain spectral methods, time-frequency transforms, and machine learning algorithms-extract comprehensive material information from raw Barkhausen signals. Deep learning approaches demonstrate superior performance for automated material classification and property prediction compared to traditional statistical methods. Industrial applications span manufacturing quality control, structural health monitoring, railway infrastructure assessment, and predictive maintenance strategies. Key achievements include establishing quantitative correlations between material properties and stress states, with measurement uncertainties of ±15-20 MPa for stress and ±20 HV for hardness. Emerging challenges include standardization imperatives, characterization of advanced materials, machine learning robustness, and autonomous system integration. Future developments prioritizing international standards, physics-informed neural networks, multimodal sensor fusion, and wireless monitoring networks will accelerate industrial adoption supporting safe, efficient engineering practice across diverse sectors.